Given semantic descriptions of object classes, zeroshot learning aims to accurately recognize objects of the unseen classes, from which no examples are available at the training stage, by associating them to the seen classes, from which labeled examples are provided. We propose to tackle this problem from the perspective of manifold learning. Our main idea is to align the semantic space that is derived from external information to the model space that concerns itself with recognizing visual features. To this end, we introduce a set of "phantom" object classes whose coordinates live in both the semantic space and the model space. Serving as bases in a dictionary, they can be optimized from labeled data such that the synthesized real object classifiers achieve optimal discriminative performance. We demonstrate superior accuracy of our approach over the state of the art on four benchmark datasets for zero-shot learning, including the full ImageNet Fall 2011 dataset with more than 20,000 unseen classes.

3D object detection is an essential task in autonomous driving. Recent techniques excel with highly accurate detection rates, provided the 3D input data is obtained from precise but expensive LiDAR technology. Approaches based on cheaper monocular or stereo imagery data have, until now, resulted in drastically lower accuracies -a gap that is commonly attributed to poor image-based depth estimation. However, in this paper we argue that it is not the quality of the data but its representation that accounts for the majority of the difference. Taking the inner workings of convolutional neural networks into consideration, we propose to convert image-based depth maps to pseudo-LiDAR representations -essentially mimicking the LiDAR signal. With this representation we can apply different existing LiDAR-based detection algorithms. On the popular KITTI benchmark, our approach achieves impressive improvements over the existing state-of-the-art in image-based performance -raising the detection accuracy of objects within the 30m range from the previous state-of-the-art of 22% to an unprecedented 74%. At the time of submission our algorithm holds the highest entry on the KITTI 3D object detection leaderboard for stereo-image-based approaches. Our code is publicly available at https

Abstract. We propose a novel supervised learning technique for summarizing videos by automatically selecting keyframes or key subshots. Casting the task as a structured prediction problem, our main idea is to use Long Short-Term Memory (LSTM) to model the variable-range temporal dependency among video frames, so as to derive both representative and compact video summaries. The proposed model successfully accounts for the sequential structure crucial to generating meaningful video summaries, leading to state-of-the-art results on two benchmark datasets. In addition to advances in modeling techniques, we introduce a strategy to address the need for a large amount of annotated data for training complex learning approaches to summarization. There, our main idea is to exploit auxiliary annotated video summarization datasets, in spite of their heterogeneity in visual styles and contents. Specifically, we show that domain adaptation techniques can improve learning by reducing the discrepancies in the original datasets' statistical properties.

Analytical tools that have spatial resolution at the nanometre scale are indispensable for the life and physical sciences. It is desirable that these tools also permit elemental and chemical identification on a scale of 10 nm or less, with large penetration depths. A variety of techniques in X-ray imaging are currently being developed that may provide these combined capabilities. Here we report the achievement of sub-15-nm spatial resolution with a soft X-ray microscope--and a clear path to below 10 nm--using an overlay technique for zone plate fabrication. The microscope covers a spectral range from a photon energy of 250 eV (approximately 5 nm wavelength) to 1.8 keV (approximately 0.7 nm), so that primary K and L atomic resonances of elements such as C, N, O, Al, Ti, Fe, Co and Ni can be probed. This X-ray microscopy technique is therefore suitable for a wide range of studies: biological imaging in the water window; studies of wet environmental samples; studies of magnetic nanostructures with both elemental and spin-orbit sensitivity; studies that require viewing through thin windows, coatings or substrates (such as buried electronic devices in a silicon chip); and three-dimensional imaging of cryogenically fixed biological cells.

Abstract. Zero-shot learning (ZSL) methods have been studied in the unrealistic setting where test data are assumed to come from unseen classes only. In this paper, we advocate studying the problem of generalized zero-shot learning (GZSL) where the test data's class memberships are unconstrained. We show empirically that naively using the classifiers constructed by ZSL approaches does not perform well in the generalized setting. Motivated by this, we propose a simple but effective calibration method that can be used to balance two conflicting forces: recognizing data from seen classes versus those from unseen ones. We develop a performance metric to characterize such a trade-off and examine the utility of this metric in evaluating various ZSL approaches. Our analysis further shows that there is a large gap between the performance of existing approaches and an upper bound established via idealized semantic embeddings, suggesting that improving class semantic embeddings is vital to GZSL.

No abstract

Video summarization has unprecedented importance to help us digest, browse, and search today's evergrowing video collections. We propose a novel subset selection technique that leverages supervision in the form of human-created summaries to perform automatic keyframe-based video summarization. The main idea is to nonparametrically transfer summary structures from annotated videos to unseen test videos. We show how to extend our method to exploit semantic side information about the video's category/genre to guide the transfer process by those training videos semantically consistent with the test input. We also show how to generalize our method to subshot-based summarization, which not only reduces computational costs but also provides more flexible ways of defining visual similarity across subshots spanning several frames. We conduct extensive evaluation on several benchmarks and demonstrate promising results, outperforming existing methods in several settings.

Abstract:To extend soft x-ray microscopy to a resolution of order 10 nm or better, we developed a new nanofabrication process for Fresnel zone plate lenses. The new process, based on the double patterning technique, has enabled us to fabricate high quality gold zone plates with 12 nm outer zones. Testing of the zone plate with the full-field transmission x-ray microscope, XM-1, in Berkeley, showed that the lens clearly resolved 12 nm lines and spaces. This result represents a significant step towards 10 nm resolution and beyond. A. Legros, and C. A. Larabell, "High resolution protein localization using soft x-ray microscopy," J.